Likelihood Ratios Given Activity-Level Propositions for DNA Transfer Evidence: Theoretical Foundations of the HaloGen Framework (Part I)

This paper establishes the theoretical foundations of HaloGen, an open-source hierarchical Bayesian framework that evaluates trace DNA evidence under activity-level propositions by explicitly modeling transfer, persistence, and detection probabilities to provide transparent and robust likelihood ratios across diverse evidentiary scenarios.

The Gist

Imagine you are a detective trying to figure out what happened at a scene based on tiny, invisible clues left behind: DNA. Usually, detectives just ask, "Whose DNA is this?" But this paper, HaloGen, asks a much harder question: "How did this DNA get here, and what does that tell us about what people were actually doing?"

Here is the breakdown of the paper's ideas using simple language and analogies:

1. The "Ghost" Problem (Zero Transfer)

In the old way of thinking, if you find DNA, you assume the person was there. If you don't find DNA, you assume they weren't. But life isn't that simple. Sometimes a person is right there, shaking hands or hugging, but they leave no DNA behind. It's like a ghost who visits a room but leaves no footprints.

HaloGen is built to handle these "ghosts." It explicitly calculates the chance that a relevant person was there but simply didn't leave a detectable trace. It doesn't ignore the silence; it listens to it.

2. The "Pass-It-On" Game (Transfer Mechanisms)

The paper looks at how DNA moves.

• Direct Transfer: You shake hands with someone, and their DNA is on your hand.
• Secondary Transfer: You shake hands with someone, then shake hands with a third person. Now the third person has DNA from the first person, even though they never met!

HaloGen treats these activities like a game of "telephone." It models how actions (like a handshake or a hug) turn into physical evidence (DNA) and how that evidence might hop from person to person.

3. The "All-Seeing Calculator" (The Framework)

Think of HaloGen as a super-smart, transparent calculator that weighs two competing stories (propositions) about what happened:

• Story A: The suspect did the action.
• Story B: Someone else did the action, or the suspect was just a bystander.

This calculator doesn't just look at one piece of evidence. It looks at the whole puzzle:

• How many people contributed to the DNA mix?
• Are there multiple stains (clues) at different spots?
• Did some people leave DNA and others leave nothing?

It combines all these clues into one big number called a Likelihood Ratio. This number tells you: "Given all these clues, is Story A much more likely than Story B?"

4. The "Safety Brake" (The Fail-Rate)

One of the most important features of HaloGen is how it handles the "ghosts" (people who left no DNA).

If the system finds no DNA from a suspect, a naive computer might say, "Aha! The suspect definitely wasn't there!" and give a huge score against them. But that's dangerous because, as we said, people sometimes leave no trace.

HaloGen uses a "fail-rate" safety brake. It says, "Okay, we know it's possible for a person to be there and leave nothing. We won't let the math go crazy and invent a huge score just because the DNA is missing."

• If there is DNA: The system uses the amount to support the story of direct contact.
• If there is NO DNA: The system stays neutral or leans slightly toward the defense (the person who didn't do it), refusing to invent a "guilty" story just because the evidence is missing.

5. The Bottom Line

This paper is the blueprint for HaloGen. It explains the math and the logic behind how the system works. It promises that the system is:

• Transparent: You can see how it reaches its conclusions.
• Stable: It won't give wild, unpredictable answers.
• Fair: It prevents the system from making up guilt just because a clue is missing.

Note: This specific paper only builds the engine and explains the theory. It does not yet show the car driving on the road (real-world case studies); that is saved for a "Part II" paper mentioned in the abstract.

Technical Summary

Technical Summary: Theoretical Foundations of the HaloGen Framework

Problem Statement
The interpretation of trace DNA evidence at the activity level presents significant challenges, primarily due to the complexity of modeling transfer, persistence, and the critical possibility that a relevant actor may leave no detectable DNA. Current interpretations often struggle to coherently integrate these factors, particularly when dealing with scenarios involving zero transfer, multiple contributors, specified unknown contributors, unobserved actors, and multiple stains. There is a need for a framework that can explicitly model these mechanisms while handling both observed DNA quantities and non-detects within a unified probabilistic structure.

Methodology
This paper establishes the theoretical foundations of HaloGen, an open-source hierarchical Bayesian framework designed to evaluate DNA quantities under competing activity-level propositions. The methodology is characterized by the following components:

• Propositional Framework: The framework frames propositions in terms of alleged activities or actions. Direct and secondary transfer are treated as mechanisms through which these activities result in observed DNA.
• Exhaustive Likelihood-Ratio Approach: Evidence is evaluated using an exhaustive propositions likelihood-ratio framework. This approach combines information across multiple contributors and stains while propagating uncertainty regarding transfer and detection.
• Unified Handling of Detection and Non-Detection:
  • Detected Quantities: Evaluated using conditional-on-detection densities.
  • Non-Detects: The probability that a relevant actor leaves no detectable DNA is represented by an empirically constrained fail-rate parameter.
• Scenario Coverage: The model accounts for complex scenarios including zero transfer, multiple contributors, specified unknown contributors, unobserved actors, and multiple stains.

Key Contributions
The primary contribution of this work is the formalization of the HaloGen framework, which offers a transparent and stable probabilistic structure for activity-level interpretation. Key technical advancements include:

1. Integration of Zero Transfer: Explicitly modeling the possibility that a relevant actor leaves no DNA, preventing the framework from forcing a detection where none exists.
2. Empirical Fail-Rate Policy: The introduction of an empirically clamped fail-rate policy. This mechanism prevents the spurious inflation of likelihood ratios in situations where the non-detection of a relevant actor is plausible.
3. Stable Behavior: The framework is designed to yield behavior where informative DNA quantities support propositions involving direct transfer, while low-information or no-detect situations result in neutral or defense-conservative outcomes.

Results and Claims
As this paper focuses on theoretical foundations, it does not present experimental validation or specific casework results; these are reserved for a companion paper. However, the paper claims that the theoretical structure of HaloGen ensures:

• Transparency and Stability: The framework behaves predictably across different evidence scenarios.
• Conservative Handling of Uncertainty: By utilizing the fail-rate parameter, the model avoids over-interpreting non-detection as evidence of absence, thereby maintaining a defense-conservative stance when appropriate.
• Coherent Uncertainty Propagation: The hierarchical Bayesian approach successfully propagates uncertainty through the complex chain of activity, transfer, and detection.

Significance
The significance of this paper lies in its establishment of the rigorous theoretical basis required for the HaloGen framework. By defining how DNA quantities and non-detects are handled within a single probabilistic framework, the authors provide the necessary groundwork for the subsequent validation and application of the framework in real-world casework. The paper asserts that this theoretical foundation is essential for moving beyond simplistic interpretations of DNA evidence toward a more robust, activity-level evaluation that accounts for the full spectrum of transfer possibilities.